
310 Empirical Research in Software Engineering
Null hypothesis: There is no signicant difference between the performance of model
predicted using Bagging technique and the model predicted using LR technique.
Alternative hypothesis: There is a signicant difference between the performance
of model predicted using Bagging technique and the model predicted using LR
technique.
After applying paired t-test using the procedure given in Section 6.4.6.3, the t-statistic
is 3.087 (p-value = 0.037)and the test is signicant at 0.05signicance level. Hence, null
hypothesis is rejected and the alternative hypothesis is accepted. The example above dem-